By David Aaron

In trading and risk management, front-office people rarely manage to convey their understanding of risk concepts and mathematical expertise to other staff members who need to support the business. If you're one of the many jumping on the Value-at-Risk bandwagon, chances are good that you or your company can benefit from Zoologic's most recent software release, Risk Management 101.

This Windows-based training tool explains in a clear and concise manner the important topics associated with VAR, from portfolio basics (determining mark-to-market, weighting returns and volatility) to more complex matters (correlation analysis and Monte Carlo simulation).

Zoologic was formed several years ago by former employees of Bankers Trust Co., one of the early proponents of using VAR measures as a means of measuring portfolio risk. It is obvious that the insights gained from earlier experiences have been well-integrated into the software.

The three lessons in Risk Management 101 provide an accurate and concise depiction of the most important topics people might ask about VAR and portfolio pricing-how a portfolio should be valued, how the risks of different instruments get treated when they are grouped together, and how to determine the maximum expected portfolio risk with different confidence levels. Risk Management 101 demystifies these issues and describes the building blocks and intuition behind the approaches.

There are three separate lessons, each with its own set of exercises. Tests can easily be designed to cover any or all of the lessons. What makes Risk Management 101 so appealing are its flexibility, its intuitive feel and the dynamic nature in which the lessons are presented. Many of the topics include animated graphics that enliven the learning process. Other features include toggles and slide bars that allow users to change the assumptions being used in the problems. For example, it is possible to adjust volatility levels to see changes in the shape of distribution curves, or to see plots of portfolio returns with different correlation assumptions.

Each subchapter has its own set of exercises that can be answered either immediately following the lesson or as a review of the topic area. This is a clever way of allowing users to choose how they prefer to learn. Responses can be checked with a simple click of the mouse, and there is always a clear, concise explanation of each issue relevant to the question. If calculations are needed, the steps are broken down and displayed.

The built-in test creator allows users to monitor their progress and understanding of the materials without having to respond to the same questions again and again. Each request for a test generates a new set of questions and answers. Personally, I would have preferred fewer multiple choice-type questions, because it is sometimes possible to determine the correct answer without knowing the material, but the questions are well-constructed nonetheless.

Users successfully completing the course should feel comfortable that they will not be lost in their next risk meeting.

David Aaron is a director at New York-based DerivaTech Consulting, which specializes in providing pricing tools for the currency options market.

Ever since RiskMetrics was introduced more than two years ago, people have been trying to find an easy way to access and manipulate the huge amount of information stored in the downloadable matrix available on JP Morgan's web site (http://www.jpmorgan.com/riskmanagement/riskmetrics/).

The site now offers a simple on-line VAR calculator that accesses the latest version of its data (http://www.jpmorgan.com/riskmanagement/var/varcalc.html). There are also a variety of shareware utilities that help pull the data into Excel spreadsheets, as well as a variety of low- and high-priced systems that access the RiskMetrics data. (See "Low-end VAR Packages for Corporate," Tech Notes, December­January.)

Now, however, Innova Financial Solutions, a Danish R&D consultancy, has gone one step further with RMOnline, a free Internet site that performs Value-at-Risk calculations on a given portfolio, using JP Morgan's RiskMetrics methodology and the daily data sets provided by Morgan and Reuters. (See http://www.ifs.dk.)

RMOnline can be used with most standard web browsers. After logging on, users enter the base currency and data changes, including the value, instrument, asset class and maturity. Pop-up menus help the data entry and error messages quickly locate improper entries. The system supports risk-monitoring of interest rates, foreign exchange, commodity and equity index prices. Thirty-one different countries are supported, along with non-U.S. dollar portfolios.

There is no limit to the size of the portfolio users can enter, and there are also ID, password and other security features that allow users to save portfolios privately on the server for future use. Only subsequent changes in the portfolio need be entered for a fresh calculation.

Among other parameters, users need to conform to the vertices used in the RiskMetrics data sets. This guideline helps organize the portfolio around different types of market risk. For instance, because a bond has both foreign exchange and interest rate risk, its present value must be entered first as a foreign exchange position in basis currency equivalent of the foreign currency amount and then as an interest rate position distributed on the relevant RiskMetrics vertices. RMOnline allows users to solve some relatively simple "what if" scenarios.

The site is clearly meant to help support Innova's high-end consulting expertise in financial instruments and risk analysis. Along with the site's ample instructions, there are clarifications and caveats pertaining to the RiskMetrics methodology. For example, users are advised that while RMOnline is a good starting point for treasury or portfolio risk management, it should not be relied upon for measuring portfolio market risk precisely. Proper treasury or portfolio management involves the "whole organization," information flow and other factors referred to as "extended" risk management. The company also offers an intranet implementation that can be run locally, tailored to an organization's hardware, operating system, web server and the appropriate databases.

Dealers who create a new instrument on their Excel spreadsheets often have to wait several long months before their models are fully integrated into their firm's middle- and back-office systems. Systems vendors have come up with a variety of ways around this problem, but most solutions involve some sort of programming within a pre-existing systems package or tool kit. In many cases, the new products are fully integrated just as they're ready to turn into vanilla instruments complete with microscopic profit margins.

Integral Development Corp. now claims it has a new product-even a new systems development category or environment-that will dramatically shorten that time period. Its Universal Finance Server allows users to develop systems using either Integral's analytics or their own and write applications in a variety of languages, including Visual Basic, Java, C++ and even Excel.

Integral's server functions much like relational database servers, which are operating pieces of software. Developers do not need to write and compile programs to use them. If a trader wants to compute the number of days between two dates or add a commission rate to a bond, for example, he sends a message or command to the server. For more complex situations, developers can program for specific tasks by writing in C++, Visual Basic, Java or HTML, and connecting to the server, which takes care of all financial processing.

This type of system divides systems responsibilities in a new way: Integral supplies the instrument, analytic and workflow models common to all capital market business areas and modules for specific business areas. The client extends these models with proprietary innovations. Integral also supplies the necessary interfaces to connect with client applications, databases and legacy systems. The client, in turn, writes client applications and connects Integral's server to its databases and legacy systems. Each group maintains the code in his or her area of responsibility. Integral predicts that it will be responsible for about 80 percent of the code, with the client responsible for the rest.

The server also handles application development a little differently. Instead of offering a series of predefined reports, it allows users to design and generate custom reports using Microsoft Word templates, Visual Basic and other localized languages. It also supports something called "hot-swappable financial instruments," which allow users to add or change financial structures in the server while it is still running-that is, without bringing the system down to load new software.

For information see: http://www.integral.com.

Trading software has to strike a delicate balance between analytical sophistication and user accessibility. Cambridge, Mass.-based Leading Market Technologies (LMT), which offers a front-office package geared specifically toward decision support and the development of custom analytics, particularly the equity and interest rate areas, is now attempting to crank up the sophistication level for trader-accessible analytics.

LMT's latest version of EXPO product, which was released in March, is a high-powered desktop analytical engine that connects to all major industry-standard sources of real-time and historical data such as Open Bloomberg, FAME, Reuters and Bridge IDF, as well as internal data sources and Internet-based data. EXPO provides traders with tools to analyze market data, test new derivatives products and distribute deal structures and pricing algorithms to networked users. In its latest release, EXPO is adding econometrics functions to its portfolio of tools. According to LMT founder Jay Smith, "We are in the final stages of beta-testing EXPOS modules that can handle very sophisticated econometrics."

These econometrics include multivariate rolling regression, which allows users to create software predictive regression equations for securities dependent on multiple variables. It can automatically roll forward the time period in question such that, if the regression requires the last 100 days of data, the software "knows" to collect the most recent available set of data. To give traders the option of testing new pricing models under different volatility scenarios, the new version of EXPO will include functions for ARCH (autoregressive conditional heteroscedacity) and ARIMA, two popular historical simulation methods for predicting volatility. Similarly, the new EXPO includes complex Monte Carlo simulation, allowing users to incorporate volatility and correlation estimates into their "random" price generation engine. And, perhaps most critical, all this sophisticated analysis can be done from a standard desktop PC. "You typically will not need a hardware upgrade to move to the next version of EXPO," says Smith.

For more information, see http://www.LMT-EXPO.com.

The market for turnkey front-through-back-office trading systems is heating up as new players continue to enter the market. Principia, which is known for its low maintenance front-through-back-office derivatives system aimed at regional banks and corporates, has made two key changes to its system with an eye toward the future.

According to Brian Donnally, the firm's trading services manager, Principia has greatly extended its foreign exchange capabilities to allow users to decompose a wide array of vanilla and exotic currency deals into their basic building blocks, and to use this information to conduct selective hedging. For example, an exotic option could be broken down into a collection of vanilla options, and its risk could also be sliced according to, say, its vega equivalent, interest rate risk and foreign exchange risk. Furthermore, the latest version of Principia lets users use cross-currency forward prices to generate forward curves for exotic currencies such as, say, the Chinese renimbi, for which forward curves are not readily available.

Although Principia allows users to break down exotic deals by their components, exotic trade records are still stored in the Principia database as single records. According to product manager John Fry, this is a critical difference between Principia and other systems that create exotic trades by linking separate vanilla tickets, because the relationship between the "component tickets"-and, therefore, how the exotic deal should be priced-may not be immediately apparent to users other than the trader who booked the original deal. "This can result in reporting inaccuracies if the wrong model is used to value the exotic deal," says Fry. "Also, I have seen situations where, say, the back office would price every linked deal using basic models and then add the total. Most exotic securities, however, are not equal to just the sum of their parts."

Fry adds that the way Principia stores exotic trades has eased the development of its second critical systems innovation: structured confirmations. He says, "In other systems, a complex derivative such as an amortized inverse floater with embedded caps and floors in addition to period caps and floors that knocks out according to an average interest rate would generate 10 or more confirms according to its disparate components. Back-office staff would then have to spend considerable time collating these confirms and reassembling them to represent the original deal." Fry explains that it is not unusual to see a deal comprising 50 or more linked tickets, which must then all be reviewed separately before confirmation.

Fry adds that Principia has also enhanced the flexibility of its back-office system to let users specify their own formats for key confirmation data, such as counterparty name, address, trade classification and so on. Says Fry, "This sounds very simple and intuitive, but most comparable systems tie users to a rigid format for the information that appears on their confirmation reports. Therefore, many back office staff spend a great of time redoing supposedly automatic confirmations by hand." And, in today's competitive global marketplace, even a badly formatted address can cost you business. "Swiss corporations really will not deal with banks that print their addresses incorrectly on the confirm statement, says Fry. This is a serious issue."

For more information, see http://www.ttllc@ppllc.com.

One of the big bones of contention in trading systems today is exactly how much firms can accomplish using the PC environment, and at what point, exactly, they should upgrade to a more powerful, UNIX server-oriented configuration. At Hathersage Capital Management, a currency option fund manager, managers have demonstrated that quite a lot can be accomplished with the tools residing on many desktops: Inventure's FENICS option pricing program and Paradox.

According to Ron Furlong, director of operations for Hathersage, "In the past, we used FENICS as both a front- and back-office system. This was because we only wanted to input deals once, and we wanted to ensure that both the front and back office were using a consistent pricing methodology." Problems, however, occurred as Hathersage grew and took on more managed accounts. He says, "Although we might, for example, execute a $10 million trade, we would then have to break that trade up into portions for allocation across various accounts according to those accounts' risk tolerance." In FENICS, this means that even a simple straddle, allocated to 10 clients, would generate 20 trade tickets, with deals representing both sides of the option being generated for all 10 clients.

About a year ago, Furlong went to Sphere, a custom development firm founded by Philip Brittan, one of the original principals of Astrogamma before its acquisition by Inventure, to create a trade consolidator written in Borland Delphi that would allow users to input deals and then allocate those deals among the firm's clients according to their risk weightings. This module, explains Furlong, would use FENICS pricing conventions and be able to export trades into FENICS for back-office analysis. "Our trade consolidator lets us set up risk categories for each client and each type of trade," he explains. "For example, if we enter into a long dollar-yen position, our more conservative clients might have a leverage ratio of 1.5 to 1, whereas our more risk-oriented clients could have a high ratio, such as 4 to 1." Any deal entered into the system is assigned a risk weighting, explains Furlong, and then is instantly split among clients according to predefined criteria.

This system currently resides on a local area network, and trades are stored in the Microsoft Access database. So far, notes Furlong, they have had very few problems with the software and no capacity problems with Microsoft Access. "This is a very low-maintenance application that works well with our operations, and the application is flexible enough to change as FENICS continues to evolve."

For more information, see http://www.spheresoft.com.

"The only competitor we really worry about right now is Reuters," says a top executive at a major risk management software firm. Reuters' risk management arm is, in fact, moving aggressively into this space. It offers Var-Net, a risk reporting service that allows users of its Sailfish middle-office system to pump data over the Reuters intranet or over the secure public Internet. Clients can upload position data into the remote Sailfish system, and then receive a selection of preformatted reports.

Reuters' market data arm is also moving aggressively into the risk management game through its relationship with JP Morgan's RiskMetrics Group. When Reuters begins maintaining RiskMetrics data sets later this year, it will add a number of bells and whistles to the data's generation as well as increased market and instrument coverage. One persistent rumor-which Reuters would neither confirm nor deny-is that users will be able to specify any level of confidence they want for volatility and correlation data. In other words, users would be able to say they would like RiskMetrics data to an 80 percent or 97 percent confidence, whereas as current RiskMetrics data available over the Internet are only available for a 95 percent level of confidence.

Bank treasury systems, which have historically lagged behind their trading desk brethren in terms of sophisticated analytics for exotics, are catching up. Or at least Rolfe and Nolan's Lighthouse trading and risk management system for bank treasuries is moving quickly in that direction. In order to better accommodate exotic equity, foreign exchange, interest rate and commodity instruments, Rolfe and Nolan has formed an alliance with Monis Software to incorporate Monis' Optimum option pricing routines into its Lighthouse system. Now Lighthouse users will have a choice-they can use standard Lighthouse models, Optimum models or their own proprietary models. Once a model is selected at the deal capture stage, it will stay with the deal through position-keeping and revaluation, ensuring consistency through the front-, middle- and back-office system. Certainly for bank treasurers who must cope with exotic deals, this is good news indeed.

For more information, see http://www.monis.co.uk. (Rolfe and Nolan did not have an Internet address as we went to press.)

The trickle-down theory of risk management goes like this: new risk management methods and systems are born on bank trading desks. Then they trickle down to corporates and-finally-to institutional investors. The Risk Standards Working Group, a consortium of plan sponsors, money managers and custodians assembled by New York-based Capital Market Risk Advisers (http://www.cmra.com) risk, helped to kick off the trend by issuing a document recommending that institutional investors thoroughly review their risk management systems and methodologies.

Now, RogersCasey, a wholly owned subsidiary of Barra, a top software vendor targeting the institutional investor market, is going one step further. In conjunction with an advisory board made up of institutional investors and investment managers, Rog-ersCasey announced that it plans to interview the 400 top institutional investors in the United States on the topic of risk management, with an emphasis on which investment risks these firms believe are most critical and the methods they are using to quantify these risks. The report, which is due out this summer, will include detailed survey results and a comprehensive analysis on risk management trends in the institutional investor community. These results will no doubt help Barra guide its own risk systems development efforts.

For more information, see http://www.barra.com and http://www.rog-erscasey.com.